[pytonslange]

It was registered as a clinical trial (https://clinicaltrials.gov/ct2/show/NCT01383161?term=NCT0138...). If you are concerned about p-hacking and replicability, what should worry you is that the outcome measures posted during registration are rather vague, i.e. ".. show less evidence of cognitive decline (as measured with neuropsychological assessments)"

At baseline, they performed an extensive neuropsychologist test battery, including a bunch of subtests:

  -Trail Making Test A
  -WAIS-III Digit Symbol Substitution
  -WAIS-III Block Design Test
  -Rey-Osterrieth Complex Figure Test (copy)
  -Trail Making Test B
  -Stroop Interference
  -F.A.S.
  -Buschke-Fuld Selective Reminding Test [SRT]
  -Wechsler Memory Scale-3rd Edition [WMS-III] 
  -Verbal Paired Associations I
  -Benton Visual Retention Test
  -Buschke-Fuld SRT
  -Rey-Osterrieth Complex Figure Test [recall]
  -WMS-III Verbal Paired Associations II
  -Boston Naming Test
  -Animal Naming Test

The outcome measures they report are:

  -TMT-A
  -SRT
  -BVMT-R

I would be _very_ surprised if those were the only tests they performed at followup after a 18 month (expensive) clinical trial. If these measures had been specified as the only outcome measures of interest, noone would question the results. But, when it was not registered, and when no other measures are reported, not even in the supplementary, the reader is left to wonder why that is.

[hanklazard]

Excellent point. Outcome measures should have been more clearly defined in pre-registration. As it was done, p-hacking and selective reporting were still possible.

[closed]

That's a good point. After having seen much of the brain training literature (especially its early studies), this wouldn't surprise me at all.